| 29 Jun 2025 |
 emily | the former is timeless advice for any innovative project but I'm sympathetic to the latter too :p | 12:56:54 |
 hexa |
works fine as long as you're not explicitly configuring conflicting things?
Exactly that.
| 12:58:19 |
| emily | seems fine then | 12:59:46 |
| emily | I agree we should definitely gradually deprecate all the scripted networking.* stuff if we're not relying on it as a new user on-ramp then | 13:00:24 |
| emily | maybe keeping a small subset of the most basic parts of the networkd shim around as deprecated-but-works indefinitely | 13:01:00 |
 antifuchs | hey folks, sorry for kicking off such a wild discussion with (what I thought was) an innocent question... I don't want to pile more work on top of y | 14:35:09 |
| emily | it's an important discussion to have | 14:35:23 |
| emily | and stuff that has become years overdue | 14:35:35 |
| emily | scripted initrd and networking are definitely going away though, sorry :P | 14:35:44 |
| antifuchs | * hey folks, sorry for kicking off such a wild discussion with (what I thought was) an innocent question... I don't want to pile more work on top of y'all's plates, and just want to say that the way forward you all are on is great (: | 14:35:46 |
| antifuchs | (hit enter too soon, lol) | 14:35:53 |
| hexa | not sorry 😄 | 14:36:05 |
| antifuchs | don't be sorry, I am looking forward to this world | 14:36:17 |
| antifuchs | however, I don't think I will be able to get this one system on it because it's gonna be a bunch of work that I can't invest right now and the cost of failure punishing (and failure will only be apparent at runtime when it's too late) | 14:37:19 |
| antifuchs | I realize there's mayyyybe a way forward for me though, that I can evaluate with fairly reduced risk; will give that a shot later (: | 14:39:02 |
| antifuchs | * however, I don't think I will be able to get this one system on it because it's gonna be a bunch of work that I can't invest right now and the cost of failure is punishing (and failure will only be apparent at runtime when it's too late) | 14:39:12 |
| emily | it sounds like you need more VM tests | 14:51:49 |
| hexa | you can try to understand why we ship modemmanager | 14:54:47 |
| emily | ok | 14:54:55 |
| hexa | diff --git a/nixos/modules/installer/cd-dvd/installation-cd-base.nix b/nixos/modules/installer/cd-dvd/installation-cd-base.nix
index d620e5e30fdf..01cb36fb99df 100644
--- a/nixos/modules/installer/cd-dvd/installation-cd-base.nix
+++ b/nixos/modules/installer/cd-dvd/installation-cd-base.nix
@@ -52,5 +52,13 @@
programs.git.enable = lib.mkDefault true;
+ # Provide networkmanager for easy wireless configuration.
+ networking.networkmanager = {
+ enable = true;
+ enableDefaultPlugins = false;
+ };
+ networking.modemmanager.enable = lib.mkForce false;
+ networking.wireless.enable = lib.mkImageMediaOverride false;
+
system.stateVersion = lib.mkDefault lib.trivial.release;
}
diff --git a/nixos/modules/installer/cd-dvd/installation-cd-graphical-base.nix b/nixos/modules/installer/cd-dvd/installation-cd-graphical-base.nix
index f73617694961..37998c334e5f 100644
--- a/nixos/modules/installer/cd-dvd/installation-cd-graphical-base.nix
+++ b/nixos/modules/installer/cd-dvd/installation-cd-graphical-base.nix
@@ -20,10 +20,6 @@
services.xserver.enable = true;
- # Provide networkmanager for easy wireless configuration.
- networking.networkmanager.enable = true;
- networking.wireless.enable = lib.mkImageMediaOverride false;
-
# KDE complains if power management is disabled (to be precise, if
# there is no power management backend such as upower).
powerManagement.enable = true;
| 14:54:59 |
| hexa | with this diff | 14:55:02 |
| hexa | * diff --git a/nixos/modules/installer/cd-dvd/installation-cd-base.nix b/nixos/modules/installer/cd-dvd/installation-cd-base.nix
index d620e5e30fdf..01cb36fb99df 100644
--- a/nixos/modules/installer/cd-dvd/installation-cd-base.nix
+++ b/nixos/modules/installer/cd-dvd/installation-cd-base.nix
@@ -52,5 +52,13 @@
programs.git.enable = lib.mkDefault true;
+ # Provide networkmanager for easy wireless configuration.
+ networking.networkmanager = {
+ enable = true;
+ enableDefaultPlugins = false;
+ };
+ networking.modemmanager.enable = lib.mkForce false;
+ networking.wireless.enable = lib.mkImageMediaOverride false;
+
system.stateVersion = lib.mkDefault lib.trivial.release;
}
diff --git a/nixos/modules/installer/cd-dvd/installation-cd-graphical-base.nix b/nixos/modules/installer/cd-dvd/installation-cd-graphical-base.nix
index f73617694961..37998c334e5f 100644
--- a/nixos/modules/installer/cd-dvd/installation-cd-graphical-base.nix
+++ b/nixos/modules/installer/cd-dvd/installation-cd-graphical-base.nix
@@ -20,10 +20,6 @@
services.xserver.enable = true;
- # Provide networkmanager for easy wireless configuration.
- networking.networkmanager.enable = true;
- networking.wireless.enable = lib.mkImageMediaOverride false;
-
# KDE complains if power management is disabled (to be precise, if
# there is no power management backend such as upower).
powerManagement.enable = true;
| 14:55:29 |
| emily | the graphical ISO already has NM I presume | 14:55:38 |
| hexa | it has, we're moving it into the base config from there | 14:55:51 |
| emily | yeah | 14:55:56 |
| emily | NM depends on ModemManager directly. I'm not sure we should omit it (because WWAN may be the only available connectivity) unless we can't reduce the closure size sufficiently without | 14:58:39 |
| emily | I'm going to look at the nix-tree | 14:58:43 |
| hexa | we're at +38MB with that diff | 14:59:03 |
| emily | +249 MiB is from the nonsense openconnect VPN thing | 14:59:15 |
| hexa | there's probably not much more that can be saved from that stack | 14:59:20 |
